suntingting

强化学习是人工智能中策略学习的一种。本程序提供一强化学习事件Q-learning demo，并绘制图形进行形象化演示。(Reinforcement learning is a kind of artificial intelligence in the strategic study。This program provides a Q-learning demo with plotting.)

C-C算法，关联积分法求嵌入维数和时间延迟。(CC algorithm, correlation integral method the embedding dimension and time delay.)

这是使用射线跟踪法求解墙体造成误差的程序，分别有穿单面墙，平行双墙，垂直双墙的情况(This is to use ray-tracing procedure for solving the error caused by the wall, respectively, through a single wall, parallel to the double wall, vertical double-wall case)

鲁棒性自适应波束形成。利用cvx工具包提高运算效率，降低运算量。(the roubust beamforming uses cvx.)

OPTIMAZATION TOOLBOOKS

无线通信中MIMO信道的空时编码技术之V-Blast编码(Wireless communication MIMO channel coding of space-time coding technology V-Blast)

L-M(Levenberg—Marquardt)是一种非线性最优化里面最常用的方法之一，改方法能快速而准确的逼近收敛值。本文简述了改方法的操作原理和具体运用，如可以方便的用于有约束与无约束最优化的求解。本文件里面有源码和测试码以及对该方法的具体介绍PDF文件。欢迎下载。(L- M (Levenberg- Marquardt) is a kind of nonlinear optimization inside one of the most commonly used method, change method can fast and accurate approximation convergence value.This paper briefly describes the operating principle of change method and concrete application, such as easy to solve constrained and unconstrained optimization.This document there are source code and test code as well as to the PDF file the detailed introduction of the method.Welcome to download.)

先定义允许状态S={(x,y)|x=0或3,y=0,1,2,3 x=y=1,2}, 允许决策D={(u,v)|u+v=1,2},求dk<-D(k=0,1,2,…,n),使 状态Sk<-S按照状态转移律S(k+1)=S(k)+(-1)^k*dk,由初始状态 S1=[3,3]经有限步到达状态S(n+1)=[0,0].此为求解多步决策问题( First define to allow the state S = ((x, y) | x = 0 or 3, y = 0,1,2,3 x = y = 1,2), to allow the decision-making D = ((u, v) | u+ v = 1,2), seek dk < -D (k = 0,1,2, ..., n), so that state Sk < -S state transition in accordance with law S (k+1) = S (k )+ (-1) ^ k* dk, from the initial state S1 = [3,3] by the finite-step to reach the status S (n+1) = [0,0]. This is a multi-step decision-making problem solving)

多尺度几何分析contourlet工具，包括去噪演示(Contourlet geometric analysis tools, including Noise Demo)

identifacation systems